Passing Efficiency of a Low Turbulence Inlet (PELTI) Final Report to NSF Prepared by the LTI Assessment Working Group: Barry J. Huebert1, Chair Steven G. Howell1 David Covert 2 Antony Clarke1 James R. Anderson3 With Collaborating Authors Bernard G. Lafleur4 Russ Seebaugh4 James Charles Wilson4 Dave Gesler4 Darrel Baumgardner5 Byron Blomquist1 1 Department of Oceanography, University of Hawaii, Honolulu, HI 96822 2 Atmospheric Science Dept, JISAO GJ-40, University of Washington, Seattle, WA 98195 3 Mechanical and Aerospace Engineering, Arizona State University, Tempe, AZ 85287 4 Department of Engineering, University of Denver, Denver, CO 80208 5 Centro de Ciencias de la Atmósfera – UNAM, Universidad Nacional Autónoma de México, Circuito Exterior s/n, Ciudad Universitaria, 04510 México City (D.F.) Address inquiries to: [email protected] 13 September, 2000 Disclaimer: The data on which this report is based was collected in July, 2000. This report is being submitted in September, 2000. It commonly takes multiple years for investigators to process, quality-check, and publish data from flight programs of this complexity. In view of the extremely short time from data collection to report, perhaps it is reasonable for the authors to attach a “preliminary” label to the tables and plots herein. Some errors will no doubt be discovered and corrected. However, we are confident that they would not change the fundamental conclusions about the functioning of the LTI. Table of Contents 3 Executive Summary Report Summary 4 A. Introduction 5 B. Approach 6 C. LTI Modeled Performance 6 D. Observations 10 E. Conclusions 11 Summary Figures 1. Introduction 15 1.1 State of the Science 16 1.2 PELTI 2. Methods 17 2.1 Overall Approach 19 2.2 LTI Configuration 20 2.3 Location of Inlets, Plumbing, and Layout 21 2.4 Bulk Chemical Analyses 22 2.5 SEM Analyses 23 2.6 Aerodynamic Particle Size Measurements 24 2.7 FSSP-300 Observations 25 2.8 Nephelometers 3. Results 26 3.1 Flight Operations 28 3.2 LTI Aerodynamic Performance 30 3.3 Bulk Filter Data 32 3.4 SEM Data 35 3.5 APS Data 37 3.6 FSSP-300 Measurements 38 3.7 Lab Measurements of Losses in Tubing 39 3.8 Modeling of Enhancements using Fluent 42 3.9 Nephelometers Discussion and Critiques of Methods 43 4.1 Tasks Remaining 43 4.2 Critique of APS Data – D. Covert 44 4.3 Critique of Anion and Cation Data – B. Huebert 45 4.4 Sources of Error in SEM Analyses – J. Anderson 46 4.5 FSSP Data Concerns – A. Clarke 48 4.6 Net LTI Passing Efficiency – J. C. Wilson 49 4.7 What Does Particle Size Mean? – S. Howell Recommendations and Conclusions 52 5.1 Comparisons of Inlets 53 5.2 Comparisons to Ambient Concentrations 54 5.3 Recommendations 54 References 57 Figures 2 Executive Summary In July, 2000 we tested the new porous-diffuser low-turbulence inlet (LTI), developed at the University of Denver, by flying it and three other inlets on NCAR’s C-130 in the Caribbean, using both dust and sea salt as test aerosols. Aerosols were analyzed using bulk chemical analysis of ions on filters, scanning electron microscopy (SEM) of filters, TSI aerodynamic particle sizers (APSs), and FSSP-300 (300) optical particle counters. We found that the LTI consistently admitted more particles to the airplane than did either the NCAR Community Aerosol Inlet (CAI) or a shrouded solid-diffuser/curved-tube inlet (SD). APS size distributions behind the other inlets began to diverge from LTI values above 1-3 um, with mass concentrations of larger particles lower by as much as a factor of ten behind the CAI and a factor of 2 behind the SD. Modeling of particle trajectories in the LTI with Fluent predicts less than a factor of two enhancement of particles between a few and 7 microns. This was supported by the SEM analyses of particles behind the LTI and TAS. Comparisons of bulk chemistry with an external reference total aerosol sampler (TAS) found no significant differences between the LTI and TAS, but both the SD and CAI passed lower values for most of the ions analyzed. Thus, the LTI filters can be used to determine the ambient mass mixing ratios of the analyzed ions. The inertial enhancements in the LTI diffuser and estimates of losses in transport to the LTI filter must be taken into account to accurately infer ambient concentrations based on LTI sampling. When this is done, the ambient mass mixing ratios estimated from the LTI filters agree within 20% with the mixing ratios determined from the TAS filters. Relative to the LTI, the SD and CAI transmission efficiencies (the concentration in the sample flow divided by the ambient concentration) was lowest for “wet” aerosol (i.e., sea-salt), apparently because salt droplets are more likely than dry dust to stick when they impact on the walls of the other inlets. We set out to test two hypotheses: A. The LTI has a demonstrably higher aerosol sampling or transmission efficiency than both the CAI (the NCAR C-130 community aerosol inlet) and traditional solid diffusers for particles in the 1-7 um range. This hypothesis could not be falsified. All the chemical and physical evidence indicates that the LTI admits more particle mass in this range than the other inlets do. However, we note that our real goal is to achieve efficiencies near unity. The ubiquity of losses in earlier inlets lead us initially to state this hypothesis in terms of “higher” efficiency, but enhancement by the LTI may cause efficiencies substantially above 1 for particles larger than 3-5 um. Since enhancements in laminar flow are calculable, most measurements can be corrected for them. 3 B. It is possible, using the LTI, to sample and characterize the number-size and surface area-size distributions of ambient dust and seasalt inside an aircraft with enough accuracy that uncertainties arising from inlet losses will contribute less than 20% to the assessment of radiative impacts. This hypothesis also could not be falsified. It essentially asks how well aerosol size distributions behind the LTI represent ambient particle distributions and light scattering. The LTI bulk chemical concentrations were statistically identical to the TAS bulk concentrations of Na+, Cl-, = ++ ++ SO4 , Mg , and Ca , which represent the ambient mixing ratios of those species. The SEM analyses showed that the LTI and TAS number concentrations were statistically identical up to 2 mm; above 2 mm the LTI showed enhancement within model-predicted limits. When Fluent model-derived inertial enhancements associated with the LTI diffuser and losses associated with transport to the LTI filter are taken into consideration, they explain the observed 20% agreement between LTI and TAS. Corrections for modest but predictable LTI enhancements also provide light scattering assessments that are representative of ambient size distributions up to 7um. Additional contributions to scattering from yet larger aerosol are unlikely to approach 20% for realistic aerosol cases. The error in radiative forcing due to positive and negative sampling biases depends both on the transmission efficiency and the fraction of the mass and total optical depth in each size interval. For those sizes that contribute little to the optical depth, a poor transmission efficiency will cause little error in radiative forcing calcualtions. It is worth noting, however, that over- or under-sampled sizes could still cause significant errors for other issues, such as the computation of deposition fluxes and heterogeneous reaction rates. CONCLUSION: Our conclusion, therefore, is that the LTI represents a significant advance in our ability to sample populations of large particles from aircraft. Its efficiency is near enough to unity to enable defendable studies of the distributions and impacts of both mineral dust and sea salt. Corrections will need to be applied for enhancement of particles in the 3-7 um range. We recommend that the ACE-Asia program use LTIs to provide samples to the various aerosol instruments on board the NCAR C-130. Report Summary A. Introduction It has long been known that typical diffuser-and-curved-tube airborne inlet systems remove particles from sampled airstream, so that instruments downstream receive air that has been depleted of supermicron particles. Since most instruments require that air be decelerated from aircraft velocities to a few m/s prior to its analysis, decelerating diffusers have been widely used in airborne sampling. Apparently the highly-turbulent flow just inside the tip of these conical diffusers causes the largest particles to be impacted on the walls of the diffuser. With the possible exception of mineral particles that may bounce off the walls, this has the effect of removing large particles and distorting the particle-size spectrum behind diffusers. 4 A workshop was convened at NCAR in 1991 to assess the state of knowledge about inlet systems. Attendees concluded that it was not possible at that time to sample supermicron particles from aircraft without substantial and unquantifiable size-dependent negative biases, and made several recommendations for ways to study and improve airborne aerosol inlets (Baumgardner et al., 1991). The notion of a shallow angle diffuser with a shrouded inlet prompted the design of the NASA SD employed in this study. Similarly the NCAR Community Aerosol Inlet (CAI) incorporated several features intended to minimize artifacts. One of the most promising suggestions from the 19901 workshop was that of Denver University researcher Russell Seebaugh, who noted that aerodynamic engineers have for years suppressed turbulence in diffusers by using boundary layer suction to prevent the separation of the boundary layer from the diffuser walls. Since that time, Seebaugh and his colleagues Bernard LaFleur and James C. Wilson have developed that concept in the laboratory.